How Justin Ernest invested nearly $500M into hot startups without a traditional VC fund

Justin Ernest looked at the VC industry and saw a structural inefficiency hiding in plain sight: family offices and smaller institutional investors desperately wanted into the hottest AI cap tables, but the gatekeeping mechanics of traditional fund formation kept them locked out. So he skipped the fund entirely. Understanding how Justin Ernest invested nearly $500M into hot startups without a traditional VC fund tells you a lot about where capital formation is heading — and what that shift means for founders and developers across Asia.

What Happened

Ernest spent over five years at Playground Global, a deep tech venture firm, where he sharpened both his investment instincts and his fundraising relationships. When he left to build something of his own, he made a deliberate choice: don't spend 12 to 18 months raising a formal fund. That's the timeline he says new managers typically face when launching a traditional vehicle. Instead, he built Sabertooth Capital around a fundamentally different architecture.

The model works like this. Ernest uses his network to secure allocations of stock in high-profile, later-stage companies — the kind of deals that are technically available but practically inaccessible to most smaller investors. He then packages those individual deals and offers them to a captive group of roughly 30 smaller institutional investors through three legal structures: special purpose vehicles (SPVs), single-asset funds, and nominee structures. In the nominee model, Sabertooth Capital holds shares directly on behalf of the participating investors, keeping cap tables clean for the portfolio companies.

The portfolio reads like a highlight reel of the current AI moment: Anthropic, Anduril, SpaceX. These are companies that most institutional allocators know they need exposure to but can't access through conventional channels. According to the TechCrunch report by Marina Temkin, Ernest has deployed nearly $500 million across these deals — a number that would be remarkable for a first-time fund manager operating through traditional structures, let alone someone who deliberately avoided building one.

The legal and operational mechanics here matter. SPVs are not new. What's new is using them at this scale, with this level of deal selectivity, as a primary strategy rather than a supplementary one. Ernest essentially turned deal-by-deal syndication into an institutional-grade product — without the overhead, the LP reporting cycles, or the management fee drag that comes with a registered fund.

Why It Matters for Asia

Asia's venture landscape has long operated with a structural disadvantage relative to Silicon Valley: proximity to deal flow. The best AI rounds — the ones that define a decade — tend to fill fast, from warm networks, in San Francisco conference rooms or Signal chats. Asian family offices and institutional allocators with genuine appetite for AI exposure have historically struggled to participate at meaningful check sizes, or at all.

Ernest's model is a direct answer to that problem, and it's one that Asian capital has every reason to replicate. The ingredients are all present in the region: deep pools of family office wealth across Singapore, Hong Kong, Jakarta, and Seoul; a growing class of operators who've spent time at top-tier global firms and returned home with networks intact; and an accelerating cohort of AI-native startups that need patient, sophisticated capital without the governance overhead of traditional institutional LPs.

The SPV-and-nominee playbook also maps cleanly onto the regulatory environments of several Southeast Asian jurisdictions. Singapore's Variable Capital Company (VCC) framework, launched in 2020, was explicitly designed to make structures like this more efficient to operate. A manager running Ernest-style deal-by-deal syndication out of Singapore today has access to legal infrastructure that didn't exist five years ago.

What's perhaps most significant for the Asia tech ecosystem is the signal this sends about where AI investment gravity is shifting. The companies in Ernest's portfolio — Anthropic, Anduril, SpaceX — are all US-headquartered. But the capital flowing through vehicles like Sabertooth increasingly comes from global sources. As Asian AI companies scale to comparable valuations, the same access problem Ernest solved for US deals will need to be solved in reverse: connecting global capital to Asian AI cap tables efficiently, without 18-month fund formation cycles slowing everything down.

What This Means for Developers

If you're a developer or technical founder in Asia, the Ernest story looks like a financing story. It isn't, really. It's a systems design story — and that framing is more useful.

Ernest identified a coordination failure: willing capital on one side, compelling deals on the other, and a slow, expensive, trust-intensive process blocking the connection. His solution wasn't to work harder within the existing system. It was to redesign the interface between capital and opportunity using lighter-weight structures that preserved the essential value (access, trust, legal clarity) while stripping out the overhead (fund formation timelines, LP diversification requirements, management company infrastructure).

Technical founders building in the AI space — especially those building on platforms like MonstarX, Asia's AI-native dev platform — face analogous coordination problems constantly. How do you connect your product to the right data sources without rebuilding integrations from scratch every time? How do you move fast on a new AI feature without the overhead of standing up new infrastructure? The answer, structurally, looks a lot like what Ernest did: use lightweight, composable primitives that preserve the core value while eliminating unnecessary friction.

There's also a direct implication for how AI startups in Asia should think about their own fundraising. The traditional pitch-deck-to-term-sheet cycle assumes a world where information moves slowly and relationships are geographically constrained. That world is dissolving. Ernest's model works because he can move fast — he identifies an allocation, structures an SPV, and closes investors in a fraction of the time a traditional fund cycle would require. Founders who understand this can position themselves accordingly: make your deal legible to syndicate-style investors, keep your cap table clean, and don't assume that only traditional VC firms have access to meaningful capital.

For developers specifically, the AI investment boom Ernest is tapping into has a direct downstream effect on what gets built and what gets funded. The companies in his portfolio are building foundational AI infrastructure. That infrastructure shapes what APIs are available, what models are accessible, and ultimately what products Asian developers can ship. Tracking where capital concentrates in AI is, functionally, tracking where the tooling landscape is heading.

Key Takeaways

A few things are worth pulling out clearly from the Ernest story, because they each carry practical weight.

Speed is structural, not tactical. Ernest didn't move fast because he worked harder than traditional fund managers. He moved fast because he chose a structure — SPVs, single-asset funds, nominee arrangements — that is inherently faster to execute than a registered fund. Structure determines velocity. This applies to product development as much as it applies to capital formation.

Network density compounds differently than fund size. A $500M traditional fund requires a large team, a formal investment committee, quarterly LP reports, and a management company with real overhead. Ernest deployed a comparable amount of capital through a network of 30 investors using lean legal structures. The leverage came from relationship density and deal access, not from institutional scale. For Asian founders thinking about their own networks, this is a reminder that a small, high-trust network often outperforms a large, low-trust one.

The access problem is solvable — and increasingly, it's being solved. The narrative that the best AI deals are permanently locked to a handful of Sand Hill Road firms is weakening. Ernest's model is one example. The rise of secondary markets for pre-IPO equity is another. As more mechanisms emerge for routing capital efficiently to high-quality AI companies, the geographic and institutional barriers that have historically disadvantaged Asian investors and founders will continue to erode.

AI is the asset class, not a sector within one. The fact that Ernest's highest-profile portfolio companies — Anthropic, Anduril, SpaceX — span AI safety, defense tech, and aerospace tells you something important. AI is not a vertical. It's a horizontal capability that's repricing companies across every category. Investors and founders who treat it as a sector bet are thinking too narrowly. The ones who treat it as infrastructure — the way Ernest's LPs apparently do — are positioning for a longer arc.

Regulatory infrastructure is catching up. The VCC framework in Singapore, similar structures emerging in other Asian jurisdictions, and increasing regulatory clarity around SPV formation mean that the operational playbook Ernest is running in the US is increasingly executable from Asia. The gap between what's possible in San Francisco and what's possible in Singapore, in terms of capital formation mechanics, is narrower than it's ever been.

Ernest's model won't work for everyone — it runs on a specific kind of network capital that takes years to build, and the deals he accesses exist because of relationships forged over half a decade at a top-tier firm. But the underlying logic is transferable: find the coordination failure, design a lighter structure that solves it, and move before the window closes. That's as true for building AI products as it is for building investment vehicles.